Reshapes the arrays returned by back_transform into
draws_matrix objects, one per quantity, so that
the posterior and bayesplot tools can be used on them.
Arguments
- x
A
size_shape_drawsobject fromback_transform.
Value
A named list of draws_matrix objects, one per quantity of
x, with the variables named as the models report them:
G[i,j], E[i,j] and P[i,j] for every element of the
covariance matrices, corrG[i,j] and corrP[i,j] for the
correlations above the diagonal only, and h2[i] or
sigma_P[i] for the vectors. The trait names are kept in
attr(, "traits").
Details
Correlations are reported once per pair, for i < j, because
the two halves of the matrix are identical and a draws object has no
notion of symmetry.
To get everything back in a single draws object, bind the elements with
do.call(posterior::bind_draws, unname(draws)). The names of the
list have to be dropped, because bind_draws() takes the draws
objects as unnamed arguments.
Examples
ss <- size_shape(cbind(a = rnorm(20), b = rnorm(20), c = rnorm(20)))
k <- length(ss$raw_traits)
G <- random_covariance(k, min_var = 0.5, max_var = 1.5)
E <- random_covariance(k, min_var = 0.5, max_var = 1.5)
m <- cbind(matrix(rep(as.vector(G), 3), 3, byrow = TRUE),
matrix(rep(as.vector(E), 3), 3, byrow = TRUE))
colnames(m) <- c(sprintf("G[%d,%d]", rep(seq_len(k), k), rep(seq_len(k), each = k)),
sprintf("E[%d,%d]", rep(seq_len(k), k), rep(seq_len(k), each = k)))
draws <- to_draws_list(back_transform(m, ss))
names(draws)
#> [1] "G" "E" "P" "corrG" "corrP" "h2"
posterior::summarise_draws(draws$corrG)
#> # A tibble: 3 × 10
#> variable mean median sd mad q5 q95 rhat ess_bulk ess_tail
#> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 corrG[1,2] 0.496 0.496 0 0 0.496 0.496 NA NA NA
#> 2 corrG[1,3] 0.199 0.199 0 0 0.199 0.199 NA NA NA
#> 3 corrG[2,3] 0.243 0.243 0 0 0.243 0.243 NA NA NA
combined <- do.call(posterior::bind_draws, unname(draws))
posterior::variables(combined)[1:4]
#> [1] "G[1,1]" "G[2,1]" "G[3,1]" "G[1,2]"
